Effects of a conditioned reinforcer upon accuracy of match-to-sample behavior in pigeons.

Researchers worked with pigeons on a matching-to-sample task. Birds had to peck the key that looked like the sample.

An orange light flashed for half a second after every wrong choice. The flash had been paired with food earlier, so it worked like a conditioned reinforcer.

The team used an ABAB design. They added the flash, removed it, added it back, then removed it again.

When the orange flash followed errors, the pigeons got much worse at matching. Accuracy dropped sharply.

Taking the flash away fixed performance. Bringing the flash back hurt accuracy again.

Davis et al. (1994) extends this idea. They showed that a quick flash right after correct choices speeds up new learning. Timing matters: same stimulus, opposite result.

Locurto et al. (1980) and Wilkie et al. (1981) line up with the drop. Both found that extra visual events—food for errors or lights during delays—also hurt matching. The 1970 paper adds a twist: the stimulus itself was the reward.

Winett et al. (1991) gives the reason why. Only stimuli paired with food act like food. Without that history, the orange flash would have done nothing.

The study warns us that conditioned reinforcers can pull behavior the wrong way if they follow errors. In clinic, a fun sound or sticker after an incorrect response could accidentally strengthen the mistake. Check your error-correction plan: make sure reinforcement lands only on correct responses, and keep any conditioned reinforcers tied to accuracy, not to the slip-up.